Selective degeneration of a physiological subtype of spinal motor neuron in mice with SOD1-linked ALS

Abstract

Amyotrophic lateral sclerosis (ALS; Lou Gehrig's disease) affects motor neurons (MNs) in the brain and spinal cord. Understanding the pathophysiology of this condition seems crucial for therapeutic design, yet few electrophysiological studies in actively degenerating animal models have been reported. Here, we report a novel preparation of acute slices from adult mouse spinal cord, allowing visualized whole cell patch-clamp recordings of fluorescent lumbar MN cell bodies from ChAT-eGFP or superoxide dismutase 1-yellow fluorescent protein (SOD1YFP) transgenic animals up to 6 mo of age. We examined 11 intrinsic electrophysiologic properties of adult ChAT-eGFP mouse MNs and classified them into four subtypes based on these parameters. The subtypes could be principally correlated with instantaneous (initial) and steady-state firing rates. We used retrograde tracing using fluorescent dye injected into fast or slow twitch lower extremity muscle with slice recordings from the fluorescent-labeled lumbar MN cell bodies to establish that fast and slow firing MNs are connected with fast and slow twitch muscle, respectively. In a G85R SOD1YFP transgenic mouse model of ALS, which becomes paralyzed by 5-6 mo, where MN cell bodies are fluorescent, enabling the same type of recording from spinal cord tissue slices, we observed that all four MN subtypes were present at 2 mo of age. At 4 mo, by which time substantial neuronal SOD1YFP aggregation and cell loss has occurred and symptoms have developed, one of the fast firing subtypes that innvervates fast twitch muscle was lost. These results begin to describe an order of the pathophysiologic events in ALS.

Keywords: amyotrophic lateral sclerosis; electrophysiology; motor neurons; neurodegeneration.

Fig. 1.

Imaging and whole cell patch-clamp recording from lumbar spinal cord MNs in acute slices. Acute spinal cord slices were prepared from a 3-mo-old ChAT-eGFP mouse as described in Methods. (A, Left) Ventral horn of a slice (red box) under DIC imaging at 4× magnification (Upper) and same ventral horn under fluorescence microscopy with a GFP/YFP filter set (Lower). Individual brightly GFP-fluorescent MNs are readily observable in the ventral horn. (Right) DIC and fluorescence imaging at 60× magnification. A single large fluorescent MN, ∼20 μm in length along a side of the pyramid and diffusely fluorescent, is readily visible (Lower) and is also localizable by DIC imaging (Upper; red circle with arrow). These images allow for visualized whole cell patch-clamp recording. (B) Whole cell patch-clamp recording from such a neuron. Neurons recorded in current clamp showed a stable resting membrane potential (−75 mV), produced large amplitude action potentials to depolarizing current pulses (Inset Upper Right; shorter time scale), and generated negative voltage deflections to hyperpolarizing current pulses (Upper Center). Neurons also received robust synaptic inputs (Inset Upper Left; expanded y axis). Neurons responded to increasing amplitudes of current pulses by markedly increasing their firing rate (lower three panels as example). (C) Cell morphology revealed through intracellular injection of biocytin during recording and subsequent streptavadin-AlexaFluor594 staining after fixation and permeabilization (false-colored yellow; Methods). A complex dendritic pattern, left intact by the slicing procedure, is observed in the slice.

Fig. 2.

Intrinsic properties of adult mouse lumbar MNs define four subgroups (clusters) of cells. (A) Dendrogram of 11 dimensional cluster analysis using Ward’s method with a Euclidean distance measure to differentiate 42 patched cells based on the 11 intrinsic properties described in Methods. Horizontal dashed line denotes the cutoff that defines four primary clusters, with vertical dashed lines bounding them. (Inset) Amalgamation schedule for the clustering, indicating linkage distances at successive clustering steps. The relatively shallow slope after formation of cluster 4 validates the cutoff point. (B) Decision tree analysis (Methods) shows that clusters formed with analysis in A, represented by different colored bars in B, can be reproduced with >90% accuracy using only first instantaneous firing frequency (FIF) and steady-state firing frequency (SSF) to a current pulse of 3× rheobase. Numbers indicate the number of neurons in each box or column. (C) 2D scatter plot of FIF vs. SS using cells from A and B show the four clusters as determined by cutoffs in B (dashed lines). Small arrows denote the four cells misclassified in moving from an 11 to a 2D classification scheme. (D) Representative examples of spike trains fired by members of the different clusters to a 500-ms current pulse of 3× rheobase. (Insets) Instantaneous firing frequencies as a function of time during the current pulse. Both instantaneous and steady state firing rates increase moving from clusters 1–4.

Fig. 3.

Retrograde transport of injected Evans blue dye into fast or slow twitch muscle predominantly marks fast firing and slow firing MNs in the respective MN pools. (A) Schema of dye injection, retrograde transport, and labeling of MN cell bodies, followed by slice production and whole cell patch-clamp recording. TA, tibialis anterior, a fast twitch muscle that dorsiflexes the foot; S, soleus, a slow twitch muscle, mediating plantar flexion of the foot. Yellow denotes the retrograde transport of dye from the injected muscle to a cell body in the respective motor pool in the lumbar cord (note that the TA pool is lateral to the S pool, although there is some overlap of MNs). (B and C) Retrograde labeling of spinal cord MN pools by injection of Evans blue into TA (B) or soleus (C) muscle. (Left) Blue color of the injected muscle. (Center) Positions of the respective motor pools in the lumbar spinal cord (L5) according to ref. (adapted) and a 4× DIC image corresponding to the boxed region (Lower) in each. (Right) Corresponding fluorescence images for eGFP (Upper) and Evans blue (EB, false-colored red; Lower) are shown at 4× (Left) and 60× (Right). In the 60× panels, the same neuron is imaged at both fluorescence wavelengths, showing colabeling by GFP and the retrograde tracer and hence positive identification of the neuron as belonging to the respective MN pool. Further to the right, panels show examples of patch-clamp recordings from doubly fluorescent neurons of each pool. The firing frequency of the TA neuron (Upper) corresponds to cluster 3, whereas that of the soleus neuron (Lower) corresponds to cluster 2. (Scale bar: Center, 200 µm; 60× panels, 10 µm.) (D) Summary of cluster types identified by whole cell patch-clamp recording from doubly fluorescent cells labeled by TA or soleus muscle injection. The pie charts indicate the percentage of each cluster type, with the number of patched cells for each in parenthesis. Ninety-one percent of TA injection-labeled MNs were clusters 3 and 4, whereas 70% of soleus injection-labeled MNs were clusters 1 and 2. The plot below indicates the firing frequency, both instantaneous and steady state, of the MNs labeled by TA injection (red dots) or soleus injection (black dots). The soleus injection labeled mostly slow firing neurons, corresponding to clusters 1 and 2, whereas the TA injection labeled faster firing neurons, corresponding to clusters 3 and 4.

Fig. 4.

Pathology and clinical signs indicate 2–3 and 4 mo as relevant time points for recording in G85R SOD1YFP animals. (A) Ventral horns of spinal cord slices (red outlines) taken from six different G85R SOD1YFP animals: three animals at 2–3 mo of age (Upper), and three animals at 4 mo of age (Lower). The images show that aggregation of G85R SOD1YFP protein commences by 2–3 mo of age, with three animals showing a variable number of aggregates in the ventral horn of the spinal cord. At 4 mo of age, the number of aggregates in the ventral horn of three different animals is greatly reduced. (Scale bar, 200 µm.) (B, Left) YFP fluorescence from MNs of G85R SOD1YFP animals at 2 mo (Upper), showing a diffusely fluorescent MN surrounded by two protein aggregates (agg.), and at 4 mo (Lower), without visible aggregates. (Right) Recording from an MN of a 4-mo-old G85R SOD1YFP mouse. Recordings were stable, produced similar large amplitude action potentials to WT, generated by current injections (Center; Right Inset, short time scale), and received robust synaptic inputs (Left Inset). (Scale bar: 60× images, 10 µm.)

Fig. 5.

Whole cell patch clamp at 2–3 and 4 mo in G85R SOD1YFP mutant animals reveals selective hyperpolarization and loss of cluster 4 cells. (A) Distribution of cluster types in mutant animals at 2–3 and 4 mo. At 2–3 mo, the normal clustering of cells is present in mutant animals (compare Left with Fig. 2C). At 4 mo, however, there are no cluster 4 cells and a reduced number of cluster 3 cells in mutant animals. (compare with normal cluster distribution of ChAT-eGFP at 4 mo in Fig. S4) (B) Input resistance and membrane potential at different ages. Although input resistance stays largely unchanged between 2 and 4 mo in mutant cells, at 2–3 mo (2m), cluster 4 mutant cells are selectively hyperpolarized relative to those from ChAT-eGFP (right graph). At 4 mo, the remaining cluster 3 cells, as well as cluster 2 cells, are hyperpolarized relative both to control and to 2- to 3-mo-old mutant cells. *P < 0.05; **P < 0.01; ***P < 0.001; Student t test.